*/
#include "anosimcommand.h"
-#include "sharedutilities.h"
-#include "sharedsobscollectsummary.h"
-#include "sharedchao1.h"
-#include "sharedace.h"
-#include "sharednseqs.h"
-#include "sharedjabund.h"
-#include "sharedsorabund.h"
-#include "sharedjclass.h"
-#include "sharedsorclass.h"
-#include "sharedjest.h"
-#include "sharedsorest.h"
-#include "sharedthetayc.h"
-#include "sharedthetan.h"
-#include "sharedkstest.h"
-#include "whittaker.h"
-#include "sharedochiai.h"
-#include "sharedanderbergs.h"
-#include "sharedkulczynski.h"
-#include "sharedkulczynskicody.h"
-#include "sharedlennon.h"
-#include "sharedmorisitahorn.h"
-#include "sharedbraycurtis.h"
-#include "sharedjackknife.h"
-#include "whittaker.h"
-#include "odum.h"
-#include "canberra.h"
-#include "structeuclidean.h"
-#include "structchord.h"
-#include "hellinger.h"
-#include "manhattan.h"
-#include "structpearson.h"
-#include "soergel.h"
-#include "spearman.h"
-#include "structkulczynski.h"
-#include "structchi2.h"
-#include "speciesprofile.h"
-#include "hamming.h"
-#include "gower.h"
-#include "memchi2.h"
-#include "memchord.h"
-#include "memeuclidean.h"
-#include "mempearson.h"
+#include "inputdata.h"
+#include "readphylipvector.h"
//**********************************************************************************************************************
-vector<string> AnosimCommand::getValidParameters(){
+vector<string> AnosimCommand::setParameters(){
try {
- string Array[] = {"groups","label","outputdir","iters","phylip","design","sets","processors","inputdir"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
+ CommandParameter pdesign("design", "InputTypes", "", "", "none", "none", "none","anosim",false,true,true); parameters.push_back(pdesign);
+ CommandParameter pphylip("phylip", "InputTypes", "", "", "none", "none", "none","anosim",false,true,true); parameters.push_back(pphylip);
+ CommandParameter piters("iters", "Number", "", "1000", "", "", "","",false,false); parameters.push_back(piters);
+ CommandParameter palpha("alpha", "Number", "", "0.05", "", "", "","",false,false); parameters.push_back(palpha);
+ CommandParameter pinputdir("inputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(pinputdir);
+ CommandParameter poutputdir("outputdir", "String", "", "", "", "", "","",false,false); parameters.push_back(poutputdir);
+
+ vector<string> myArray;
+ for (int i = 0; i < parameters.size(); i++) { myArray.push_back(parameters[i].name); }
return myArray;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "getValidParameters");
+ m->errorOut(e, "AnosimCommand", "setParameters");
exit(1);
}
}
//**********************************************************************************************************************
-AnosimCommand::AnosimCommand(){
+string AnosimCommand::getHelpString(){
try {
- abort = true; calledHelp = true;
- vector<string> tempOutNames;
- outputTypes["anosim"] = tempOutNames;
+ string helpString = "";
+ helpString += "Referenced: Clarke, K. R. (1993). Non-parametric multivariate analysis of changes in community structure. _Australian Journal of Ecology_ 18, 117-143.\n";
+ helpString += "The anosim command outputs a .anosim file. \n";
+ helpString += "The anosim command parameters are phylip, iters, and alpha. The phylip and design parameters are required, unless you have valid current files.\n";
+ helpString += "The design parameter allows you to assign your samples to groups when you are running anosim. It is required. \n";
+ helpString += "The design file looks like the group file. It is a 2 column tab delimited file, where the first column is the sample name and the second column is the group the sample belongs to.\n";
+ helpString += "The iters parameter allows you to set number of randomization for the P value. The default is 1000. \n";
+ helpString += "The anosim command should be in the following format: anosim(phylip=file.dist, design=file.design).\n";
+ helpString += "Note: No spaces between parameter labels (i.e. iters), '=' and parameters (i.e. 1000).\n";
+ return helpString;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "AnosimCommand");
+ m->errorOut(e, "AnosimCommand", "getHelpString");
exit(1);
}
}
//**********************************************************************************************************************
-vector<string> AnosimCommand::getRequiredParameters(){
- try {
- string Array[] = {"design"};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
- }
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "getRequiredParameters");
- exit(1);
- }
+string AnosimCommand::getOutputPattern(string type) {
+ try {
+ string pattern = "";
+
+ if (type == "anosim") { pattern = "[filename],anosim"; } //makes file like: amazon.align
+ else { m->mothurOut("[ERROR]: No definition for type " + type + " output pattern.\n"); m->control_pressed = true; }
+
+ return pattern;
+ }
+ catch(exception& e) {
+ m->errorOut(e, "AnosimCommand", "getOutputPattern");
+ exit(1);
+ }
}
//**********************************************************************************************************************
-vector<string> AnosimCommand::getRequiredFiles(){
+AnosimCommand::AnosimCommand(){
try {
- string Array[] = {};
- vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
- return myArray;
+ abort = true; calledHelp = true;
+ setParameters();
+ vector<string> tempOutNames;
+ outputTypes["anosim"] = tempOutNames;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "getRequiredFiles");
+ m->errorOut(e, "AnosimCommand", "AnosimCommand");
exit(1);
}
}
AnosimCommand::AnosimCommand(string option) {
try {
- globaldata = GlobalData::getInstance();
abort = false; calledHelp = false;
- allLines = 1;
- labels.clear();
//allow user to run help
if(option == "help") { help(); abort = true; calledHelp = true; }
+ else if(option == "citation") { citation(); abort = true; calledHelp = true;}
else {
- //valid paramters for this command
- string AlignArray[] = {"groups","label","outputdir","iters","phylip","design","sets","processors","inputdir"};
- vector<string> myArray (AlignArray, AlignArray+(sizeof(AlignArray)/sizeof(string)));
+ vector<string> myArray = setParameters();
OptionParser parser(option);
map<string,string> parameters = parser.getParameters();
}
}
- phylipfile = validParameter.validFile(parameters, "phylip", true);
- if (phylipfile == "not open") { phylipfile = ""; abort = true; }
- else if (phylipfile == "not found") { phylipfile = ""; }
- else { globaldata->newRead(); format = "phylip"; globaldata->setPhylipFile(phylipfile); }
+ phylipFileName = validParameter.validFile(parameters, "phylip", true);
+ if (phylipFileName == "not open") { phylipFileName = ""; abort = true; }
+ else if (phylipFileName == "not found") {
+ //if there is a current phylip file, use it
+ phylipFileName = m->getPhylipFile();
+ if (phylipFileName != "") { m->mothurOut("Using " + phylipFileName + " as input file for the phylip parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current phylip file and the phylip parameter is required."); m->mothurOutEndLine(); abort = true; }
+
+ }else { m->setPhylipFile(phylipFileName); }
//check for required parameters
- designfile = validParameter.validFile(parameters, "design", true);
- if (designfile == "not open") { abort = true; }
- else if (designfile == "not found") { designfile = ""; m->mothurOut("You must provide an design file."); m->mothurOutEndLine(); abort = true; }
-
- //make sure the user has already run the read.otu command
- if ((globaldata->getSharedFile() == "")) {
- if ((phylipfile == "")) {
- m->mothurOut("You must read a list and a group, a shared file or provide a distance matrix before you can use the anosim command."); m->mothurOutEndLine(); abort = true;
- }
- }else { sharedfile = globaldata->getSharedFile(); }
-
- //use distance matrix if one is provided
- if ((sharedfile != "") && (phylipfile != "")) { sharedfile = ""; }
-
- //check for optional parameter and set defaults
- // ...at some point should added some additional type checking...
- label = validParameter.validFile(parameters, "label", false);
- if (label == "not found") { label = ""; }
- else {
- if(label != "all") { m->splitAtDash(label, labels); allLines = 0; }
- else { allLines = 1; }
- }
-
- //if the user has not specified any labels use the ones from read.otu
- if (label == "") {
- allLines = globaldata->allLines;
- labels = globaldata->labels;
- }
-
- groups = validParameter.validFile(parameters, "groups", false);
- if (groups == "not found") { groups = ""; }
- else {
- m->splitAtDash(groups, Groups);
- globaldata->Groups = Groups;
- }
-
- sets = validParameter.validFile(parameters, "sets", false);
- if (sets == "not found") { sets = ""; pickedGroups = false; }
- else {
- pickedGroups = true;
- m->splitAtDash(sets, Sets);
- }
-
-
- string temp = validParameter.validFile(parameters, "iters", false); if (temp == "not found") { temp = "1000"; }
- convert(temp, iters);
-
- temp = validParameter.validFile(parameters, "processors", false); if (temp == "not found"){ temp = "1"; }
- convert(temp, processors);
-
- vector<string> Estimators;
- calc = validParameter.validFile(parameters, "calc", false);
- if (calc == "not found") { calc = "morisitahorn"; }
- m->splitAtDash(calc, Estimators);
-
- if (abort == false) {
-
- ValidCalculators* validCalculator = new ValidCalculators();
-
- for (int i=0; i<Estimators.size(); i++) {
- if (validCalculator->isValidCalculator("treegroup", Estimators[i]) == true) {
- if (Estimators[i] == "sharedsobs") {
- calculators.push_back(new SharedSobsCS());
- }else if (Estimators[i] == "sharedchao") {
- calculators.push_back(new SharedChao1());
- }else if (Estimators[i] == "sharedace") {
- calculators.push_back(new SharedAce());
- }else if (Estimators[i] == "jabund") {
- calculators.push_back(new JAbund());
- }else if (Estimators[i] == "sorabund") {
- calculators.push_back(new SorAbund());
- }else if (Estimators[i] == "jclass") {
- calculators.push_back(new Jclass());
- }else if (Estimators[i] == "sorclass") {
- calculators.push_back(new SorClass());
- }else if (Estimators[i] == "jest") {
- calculators.push_back(new Jest());
- }else if (Estimators[i] == "sorest") {
- calculators.push_back(new SorEst());
- }else if (Estimators[i] == "thetayc") {
- calculators.push_back(new ThetaYC());
- }else if (Estimators[i] == "thetan") {
- calculators.push_back(new ThetaN());
- }else if (Estimators[i] == "kstest") {
- calculators.push_back(new KSTest());
- }else if (Estimators[i] == "sharednseqs") {
- calculators.push_back(new SharedNSeqs());
- }else if (Estimators[i] == "ochiai") {
- calculators.push_back(new Ochiai());
- }else if (Estimators[i] == "anderberg") {
- calculators.push_back(new Anderberg());
- }else if (Estimators[i] == "kulczynski") {
- calculators.push_back(new Kulczynski());
- }else if (Estimators[i] == "kulczynskicody") {
- calculators.push_back(new KulczynskiCody());
- }else if (Estimators[i] == "lennon") {
- calculators.push_back(new Lennon());
- }else if (Estimators[i] == "morisitahorn") {
- calculators.push_back(new MorHorn());
- }else if (Estimators[i] == "braycurtis") {
- calculators.push_back(new BrayCurtis());
- }else if (Estimators[i] == "whittaker") {
- calculators.push_back(new Whittaker());
- }else if (Estimators[i] == "odum") {
- calculators.push_back(new Odum());
- }else if (Estimators[i] == "canberra") {
- calculators.push_back(new Canberra());
- }else if (Estimators[i] == "structeuclidean") {
- calculators.push_back(new StructEuclidean());
- }else if (Estimators[i] == "structchord") {
- calculators.push_back(new StructChord());
- }else if (Estimators[i] == "hellinger") {
- calculators.push_back(new Hellinger());
- }else if (Estimators[i] == "manhattan") {
- calculators.push_back(new Manhattan());
- }else if (Estimators[i] == "structpearson") {
- calculators.push_back(new StructPearson());
- }else if (Estimators[i] == "soergel") {
- calculators.push_back(new Soergel());
- }else if (Estimators[i] == "spearman") {
- calculators.push_back(new Spearman());
- }else if (Estimators[i] == "structkulczynski") {
- calculators.push_back(new StructKulczynski());
- }else if (Estimators[i] == "speciesprofile") {
- calculators.push_back(new SpeciesProfile());
- }else if (Estimators[i] == "hamming") {
- calculators.push_back(new Hamming());
- }else if (Estimators[i] == "structchi2") {
- calculators.push_back(new StructChi2());
- }else if (Estimators[i] == "gower") {
- calculators.push_back(new Gower());
- }else if (Estimators[i] == "memchi2") {
- calculators.push_back(new MemChi2());
- }else if (Estimators[i] == "memchord") {
- calculators.push_back(new MemChord());
- }else if (Estimators[i] == "memeuclidean") {
- calculators.push_back(new MemEuclidean());
- }else if (Estimators[i] == "mempearson") {
- calculators.push_back(new MemPearson());
- }
- }
- }
- }
+ designFileName = validParameter.validFile(parameters, "design", true);
+ if (designFileName == "not open") { designFileName = ""; abort = true; }
+ else if (designFileName == "not found") {
+ //if there is a current design file, use it
+ designFileName = m->getDesignFile();
+ if (designFileName != "") { m->mothurOut("Using " + designFileName + " as input file for the design parameter."); m->mothurOutEndLine(); }
+ else { m->mothurOut("You have no current design file and the design parameter is required."); m->mothurOutEndLine(); abort = true; }
+ }else { m->setDesignFile(designFileName); }
+
+ string temp = validParameter.validFile(parameters, "iters", false);
+ if (temp == "not found") { temp = "1000"; }
+ m->mothurConvert(temp, iters);
+
+ temp = validParameter.validFile(parameters, "alpha", false);
+ if (temp == "not found") { temp = "0.05"; }
+ m->mothurConvert(temp, experimentwiseAlpha);
}
}
exit(1);
}
}
-
-//**********************************************************************************************************************
-
-void AnosimCommand::help(){
- try {
- m->mothurOut("The anosim command can only be executed after a successful read.otu command of a list and group or shared file, or by providing a phylip formatted distance matrix.\n");
- m->mothurOut("The anosim command outputs a .anosim file. \n");
- m->mothurOut("The anosim command parameters are phylip, iters, groups, label, design, sets and processors. The design parameter is required.\n");
- m->mothurOut("The design parameter allows you to assign your groups to sets when you are running anosim. It is required. \n");
- m->mothurOut("The design file looks like the group file. It is a 2 column tab delimited file, where the first column is the group name and the second column is the set the group belongs to.\n");
- m->mothurOut("The sets parameter allows you to specify which of the sets in your designfile you would like to analyze. The set names are separated by dashes. THe default is all sets in the designfile. To run the pairwise comparisons of all sets use sets=all.\n");
- m->mothurOut("The iters parameter allows you to set number of randomization for the P value. The default is 1000. \n");
- m->mothurOut("The groups parameter allows you to specify which of the groups you would like included. The group names are separated by dashes. groups=all will find all pairwise comparisons. \n");
- m->mothurOut("The label parameter allows you to select what distance levels you would like, and are also separated by dashes.\n");
- m->mothurOut("The processors parameter allows you to specify how many processors you would like to use. The default is 1. \n");
- m->mothurOut("The anosim command should be in the following format: anosim(design=yourDesignFile).\n");
- m->mothurOut("Example anosim(design=temp.design, groups=A-B-C).\n");
- m->mothurOut("Note: No spaces between parameter labels (i.e. groups), '=' and parameters (i.e.yourGroups).\n\n");
-
- }
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "help");
- exit(1);
- }
-}
-
-//**********************************************************************************************************************
-
-AnosimCommand::~AnosimCommand(){}
-
//**********************************************************************************************************************
-
int AnosimCommand::execute(){
try {
if (abort == true) { if (calledHelp) { return 0; } return 2; }
//read design file
- designMap = new GroupMap(designfile);
+ designMap = new GroupMap(designFileName);
designMap->readDesignMap();
- //make sure sets are all in designMap
- SharedUtil* util = new SharedUtil();
- util->setGroups(Sets, designMap->namesOfGroups);
- delete util;
-
- //if the user pickedGroups or set sets=all, then we want to figure out how to split the pairwise comparison between processors
- int numGroups = Sets.size();
- if (pickedGroups) {
- for (int a=0; a<numGroups; a++) {
- for (int l = 0; l < a; l++) {
- vector<string> groups; groups.push_back(Sets[a]); groups.push_back(Sets[l]);
- namesOfGroupCombos.push_back(groups);
- }
- }
- }else { //one giant compare
- vector<string> groups;
- for (int a=0; a<numGroups; a++) { groups.push_back(Sets[a]); }
- namesOfGroupCombos.push_back(groups);
- }
+ if (outputDir == "") { outputDir = m->hasPath(phylipFileName); }
- //only 1 combo
- if (numGroups == 2) { processors = 1; }
- else if (numGroups < 2) { m->mothurOut("Not enough sets, I need at least 2 valid sets. Unable to complete command."); m->mothurOutEndLine(); m->control_pressed = true; }
+ //read in distance matrix and square it
+ ReadPhylipVector readMatrix(phylipFileName);
+ vector<string> sampleNames = readMatrix.read(distanceMatrix);
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors != 1){
- int numPairs = namesOfGroupCombos.size();
- int numPairsPerProcessor = numPairs / processors;
-
- for (int i = 0; i < processors; i++) {
- int startPos = i * numPairsPerProcessor;
- if(i == processors - 1){
- numPairsPerProcessor = numPairs - i * numPairsPerProcessor;
- }
- lines.push_back(linePair(startPos, numPairsPerProcessor));
+ for(int i=0;i<distanceMatrix.size();i++){
+ for(int j=0;j<i;j++){
+ distanceMatrix[i][j] *= distanceMatrix[i][j];
}
}
-#endif
- if (sharedfile != "") { //create distance matrix for each label
-
- if (outputDir == "") { outputDir = m->hasPath(sharedfile); }
+ //link designMap to rows/columns in distance matrix
+ map<string, vector<int> > origGroupSampleMap;
+ for(int i=0;i<sampleNames.size();i++){
+ string group = designMap->getGroup(sampleNames[i]);
- //for each calculator
- for(int i = 0 ; i < calculators.size(); i++) {
-
- //create a new filename
- ofstream out;
- string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".anosim";
- m->openOutputFile(outputFile, out);
- outputNames.push_back(outputFile); outputTypes["anosim"].push_back(outputFile);
-
- //print headers
- out << "label\tgroupsCompared\tRValue\tpValue" << endl;
- m->mothurOut("label\tgroupsCompared\tRValue\tpValue"); m->mothurOutEndLine();
- out.close();
- }
-
- InputData input(sharedfile, "sharedfile");
- vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
- string lastLabel = lookup[0]->getLabel();
-
- //if the users enters label "0.06" and there is no "0.06" in their file use the next lowest label.
- set<string> processedLabels;
- set<string> userLabels = labels;
-
- //sanity check between shared file groups and design file groups
- for (int i = 0; i < lookup.size(); i++) {
- string group = designMap->getGroup(lookup[i]->getGroup());
-
- if (group == "not found") { m->control_pressed = true; m->mothurOut("[ERROR]: " + lookup[i]->getGroup() + " is not in your design file, please correct."); m->mothurOutEndLine(); }
- }
-
- //as long as you are not at the end of the file or done wih the lines you want
- while((lookup[0] != NULL) && ((allLines == 1) || (userLabels.size() != 0))) {
-
- if (m->control_pressed) { for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
- if(allLines == 1 || labels.count(lookup[0]->getLabel()) == 1){
-
- m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
- process(lookup);
-
- processedLabels.insert(lookup[0]->getLabel());
- userLabels.erase(lookup[0]->getLabel());
- }
-
- if ((m->anyLabelsToProcess(lookup[0]->getLabel(), userLabels, "") == true) && (processedLabels.count(lastLabel) != 1)) {
- string saveLabel = lookup[0]->getLabel();
-
- for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
- lookup = input.getSharedRAbundVectors(lastLabel);
- m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
-
- process(lookup);
+ if (group == "not found") {
+ m->mothurOut("[ERROR]: " + sampleNames[i] + " is not in your design file, please correct."); m->mothurOutEndLine(); m->control_pressed = true;
+ }else { origGroupSampleMap[group].push_back(i); }
+ }
+ int numGroups = origGroupSampleMap.size();
+
+ if (m->control_pressed) { delete designMap; return 0; }
+
+ //create a new filename
+ ofstream ANOSIMFile;
+ map<string, string> variables; variables["[filename]"] = outputDir + m->getRootName(m->getSimpleName(phylipFileName));
+ string ANOSIMFileName = getOutputFileName("anosim", variables);
+
+ m->openOutputFile(ANOSIMFileName, ANOSIMFile);
+ outputNames.push_back(ANOSIMFileName); outputTypes["anosim"].push_back(ANOSIMFileName);
+ m->mothurOut("\ncomparison\tR-value\tP-value\n");
+ ANOSIMFile << "comparison\tR-value\tP-value\n";
+
+
+ double fullANOSIMPValue = runANOSIM(ANOSIMFile, distanceMatrix, origGroupSampleMap, experimentwiseAlpha);
+
+
+ if(fullANOSIMPValue <= experimentwiseAlpha && numGroups > 2){
+
+ int numCombos = numGroups * (numGroups-1) / 2;
+ double pairwiseAlpha = experimentwiseAlpha / (double) numCombos;
+
+ for(map<string, vector<int> >::iterator itA=origGroupSampleMap.begin();itA!=origGroupSampleMap.end();itA++){
+ map<string, vector<int> >::iterator itB = itA;
+ itB++;
+ for(itB;itB!=origGroupSampleMap.end();itB++){
- processedLabels.insert(lookup[0]->getLabel());
- userLabels.erase(lookup[0]->getLabel());
+ map<string, vector<int> > subGroupSampleMap;
- //restore real lastlabel to save below
- lookup[0]->setLabel(saveLabel);
- }
-
- lastLabel = lookup[0]->getLabel();
- //prevent memory leak
- for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; lookup[i] = NULL; }
-
- if (m->control_pressed) { globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
- //get next line to process
- lookup = input.getSharedRAbundVectors();
- }
-
- if (m->control_pressed) { globaldata->Groups.clear(); delete designMap; for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
- //output error messages about any remaining user labels
- set<string>::iterator it;
- bool needToRun = false;
- for (it = userLabels.begin(); it != userLabels.end(); it++) {
- m->mothurOut("Your file does not include the label " + *it);
- if (processedLabels.count(lastLabel) != 1) {
- m->mothurOut(". I will use " + lastLabel + "."); m->mothurOutEndLine();
- needToRun = true;
- }else {
- m->mothurOut(". Please refer to " + lastLabel + "."); m->mothurOutEndLine();
- }
- }
-
- //run last label if you need to
- if (needToRun == true) {
- for (int i = 0; i < lookup.size(); i++) { if (lookup[i] != NULL) { delete lookup[i]; } }
- lookup = input.getSharedRAbundVectors(lastLabel);
-
- m->mothurOut(lookup[0]->getLabel()); m->mothurOutEndLine();
-
- process(lookup);
-
- for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; }
- }
-
- //reset groups parameter
- globaldata->Groups.clear();
+ subGroupSampleMap[itA->first] = itA->second; string groupA = itA->first;
+ subGroupSampleMap[itB->first] = itB->second; string groupB = itB->first;
-
- }else { //user provided distance matrix
-
- if (outputDir == "") { outputDir = m->hasPath(phylipfile); }
-
- //create a new filename
- ofstream out;
- string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "anosim";
- m->openOutputFile(outputFile, out);
- outputNames.push_back(outputFile); outputTypes["anosim"].push_back(outputFile);
-
- //print headers
- out << "groupsCompared\tRValue\tpValue" << endl;
- m->mothurOut("groupsCompared\tRValue\tpValue"); m->mothurOutEndLine();
- out.close();
-
- ReadPhylipVector readMatrix(phylipfile);
- vector< vector<double> > completeMatrix;
- vector<string> names = readMatrix.read(completeMatrix);
-
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors == 1){
- driver(0, namesOfGroupCombos.size(), names, "", completeMatrix);
- }else{
- int process = 1;
- vector<int> processIDS;
-
- //loop through and create all the processes you want
- while (process != processors) {
- int pid = fork();
+ vector<int> subIndices;
+ for(map<string, vector<int> >::iterator it=subGroupSampleMap.begin();it!=subGroupSampleMap.end();it++){
+ subIndices.insert(subIndices.end(), it->second.begin(), it->second.end());
+ }
+ int subNumSamples = subIndices.size();
+
+ sort(subIndices.begin(), subIndices.end());
- if (pid > 0) {
- processIDS.push_back(pid);
- process++;
- }else if (pid == 0){
- driver(lines[process].start, lines[process].num, names, toString(getpid()), completeMatrix);
- exit(0);
- }else {
- m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
- for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
- exit(0);
+ vector<vector<double> > subDistMatrix(distanceMatrix.size());
+ for(int i=0;i<distanceMatrix.size();i++){
+ subDistMatrix[i].assign(distanceMatrix.size(), -1);
}
+
+ for(int i=0;i<subNumSamples;i++){
+ for(int j=0;j<i;j++){
+ subDistMatrix[subIndices[i]][subIndices[j]] = distanceMatrix[subIndices[i]][subIndices[j]];
+ }
+ }
+
+ runANOSIM(ANOSIMFile, subDistMatrix, subGroupSampleMap, pairwiseAlpha);
+
}
-
- //do my part
- driver(lines[0].start, lines[0].num, names, "", completeMatrix);
-
- //force parent to wait until all the processes are done
- for (int i=0;i<(processors-1);i++) {
- int temp = processIDS[i];
- wait(&temp);
- }
-
- //append files
- string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "anosim";
- for (int j = 0; j < processIDS.size(); j++) {
- m->appendFiles((outputFile + toString(processIDS[j])), outputFile);
- remove((outputFile + toString(processIDS[j])).c_str());
- }
-
}
-#else
- driver(0, namesOfGroupCombos.size(), names, "", completeMatrix);
-#endif
+ m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
+ m->mothurOut("Pair-wise error rate (Bonferroni): " + toString(pairwiseAlpha) + '\n');
+ }
+ else{
+ m->mothurOut("\nExperiment-wise error rate: " + toString(experimentwiseAlpha) + '\n');
}
+ m->mothurOut("If you have borderline P-values, you should try increasing the number of iterations\n");
+ ANOSIMFile.close();
+
delete designMap;
-
- if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) { remove(outputNames[i].c_str()); } return 0; }
-
+
m->mothurOutEndLine();
m->mothurOut("Output File Names: "); m->mothurOutEndLine();
for (int i = 0; i < outputNames.size(); i++) { m->mothurOut(outputNames[i]); m->mothurOutEndLine(); }
}
//**********************************************************************************************************************
-int AnosimCommand::process(vector<SharedRAbundVector*> thisLookup) {
- try{
+double AnosimCommand::runANOSIM(ofstream& ANOSIMFile, vector<vector<double> > dMatrix, map<string, vector<int> > groupSampleMap, double alpha) {
+ try {
+
-#if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
- if(processors == 1){
- driver(0, namesOfGroupCombos.size(), thisLookup, "");
- }else{
- int process = 1;
- vector<int> processIDS;
-
- //loop through and create all the processes you want
- while (process != processors) {
- int pid = fork();
-
- if (pid > 0) {
- processIDS.push_back(pid);
- process++;
- }else if (pid == 0){
- driver(lines[process].start, lines[process].num, thisLookup, toString(getpid()));
- exit(0);
- }else {
- m->mothurOut("[ERROR]: unable to spawn the necessary processes."); m->mothurOutEndLine();
- for (int i = 0; i < processIDS.size(); i++) { kill (processIDS[i], SIGINT); }
- exit(0);
- }
- }
-
- //do my part
- driver(lines[0].start, lines[0].num, thisLookup, "");
-
- //force parent to wait until all the processes are done
- for (int i=0;i<(processors-1);i++) {
- int temp = processIDS[i];
- wait(&temp);
- }
-
- //append files
- for(int i = 0 ; i < calculators.size(); i++) {
- string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".anosim";
-
- for (int j = 0; j < processIDS.size(); j++) {
- m->appendFiles((outputFile + toString(processIDS[j])), outputFile);
- remove((outputFile + toString(processIDS[j])).c_str());
- }
- }
+ vector<vector<double> > rankMatrix = convertToRanks(dMatrix);
+ double RValue = calcR(rankMatrix, groupSampleMap);
+
+ int pCount = 0;
+ for(int i=0;i<iters;i++){
+ map<string, vector<int> > randGroupSampleMap = getRandomizedGroups(groupSampleMap);
+ double RValueRand = calcR(rankMatrix, randGroupSampleMap);
+ if(RValue <= RValueRand){ pCount++; }
}
-#else
- driver(0, namesOfGroupCombos.size(), thisLookUp, "");
-#endif
+
+ double pValue = (double)pCount / (double) iters;
+ string pString = "";
+ if(pValue < 1/(double)iters){ pString = '<' + toString(1/(double)iters); }
+ else { pString = toString(pValue); }
- return 0;
+
+ map<string, vector<int> >::iterator it=groupSampleMap.begin();
+ m->mothurOut(it->first);
+ ANOSIMFile << it->first;
+ it++;
+ for(it;it!=groupSampleMap.end();it++){
+ m->mothurOut('-' + it->first);
+ ANOSIMFile << '-' << it->first;
+
+ }
+ m->mothurOut('\t' + toString(RValue) + '\t' + pString);
+ ANOSIMFile << '\t' << RValue << '\t' << pString;
+
+ if(pValue < alpha){
+ ANOSIMFile << "*";
+ m->mothurOut("*");
+ }
+ ANOSIMFile << endl;
+ m->mothurOutEndLine();
+
+ return pValue;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "process");
+ m->errorOut(e, "AnosimCommand", "calcAnisom");
exit(1);
}
}
+
//**********************************************************************************************************************
-int AnosimCommand::driver(int start, int num, vector<SharedRAbundVector*> thisLookup, string pidValue) {
+double AnosimCommand::calcR(vector<vector<double> > rankMatrix, map<string, vector<int> > groupSampleMap){
try {
- vector<SharedRAbundVector*> subset;
- EstOutput data;
+
+ int numSamples = 0;
+ for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
+ numSamples += it->second.size();
+ }
- //for each calculator
- for(int i = 0 ; i < calculators.size(); i++) {
-
- //create a new filename
- ofstream out;
- string outputFile = outputDir + m->getRootName(m->getSimpleName(sharedfile)) + calculators[i]->getName() + ".anosim" + pidValue;
- m->openOutputFileAppend(outputFile, out);
- out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
- cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
-
- //for each combo
- for (int c = start; c < (start+num); c++) {
-
- if (m->control_pressed) { out.close(); return 0; }
-
- //get set names
- vector<string> setNames;
- for (int j = 0; j < namesOfGroupCombos[c].size(); j++) { setNames.push_back(namesOfGroupCombos[c][j]); }
-
- vector<SharedRAbundVector*> thisCombosLookup;
- vector<string> thisCombosLookupSets; //what set each sharedRabund is from to be used when calculating SSWithin
- for (int k = 0; k < thisLookup.size(); k++) {
- string thisGroup = thisLookup[k]->getGroup();
-
- //is this group for a set we want to compare??
- if (m->inUsersGroups(designMap->getGroup(thisGroup), setNames)) {
- thisCombosLookup.push_back(thisLookup[k]);
- thisCombosLookupSets.push_back(designMap->getGroup(thisGroup));
- }
-
- }
-
- int numGroups = thisCombosLookup.size();
-
- //calc the distance matrix
- matrix.clear();
- matrix.resize(numGroups);
- for (int k = 0; k < matrix.size(); k++) { for (int j = 0; j < matrix.size(); j++) { matrix[k].push_back(1.0); } }
-
- if (thisCombosLookup.size() == 0) {
- m->mothurOut("[ERROR]: Missing shared info for sets. Skipping comparison."); m->mothurOutEndLine();
- }else{
-
- out << thisLookup[0]->getLabel() << '\t';
- m->mothurOut(thisLookup[0]->getLabel() + "\t");
- if (setNames.size() == 2) { out << setNames[0] << "-" << setNames[1] << '\t'; m->mothurOut(setNames[0] + "-" + setNames[1] + "\t"); }
- else { out << "all" << '\t'; m->mothurOut("all\t"); }
-
- for (int k = 0; k < thisCombosLookup.size(); k++) {
- for (int l = k; l < thisCombosLookup.size(); l++) {
-
- if (m->control_pressed) { out.close(); return 0; }
-
- if (k != l) { //we dont need to similiarity of a groups to itself
- //get estimated similarity between 2 groups
- subset.clear(); //clear out old pair of sharedrabunds
- //add new pair of sharedrabunds
- subset.push_back(thisCombosLookup[k]); subset.push_back(thisCombosLookup[l]);
-
- //if this calc needs all groups to calculate the pair load all groups
- if (calculators[i]->getNeedsAll()) {
- //load subset with rest of lookup for those calcs that need everyone to calc for a pair
- for (int w = 0; w < thisCombosLookup.size(); w++) {
- if ((w != k) && (w != l)) { subset.push_back(thisCombosLookup[w]); }
- }
- }
-
- data = calculators[i]->getValues(subset); //saves the calculator outputs
-
- //save values in similarity matrix
- matrix[k][l] = 1.0 - data[0];
- matrix[l][k] = 1.0 - data[0];
- }
- }
- }
-
- //calc anosim
- calcAnosim(out, setNames.size(), thisCombosLookupSets);
+
+ double within = 0.0;
+ int numWithinComps = 0;
+
+ for(map<string, vector<int> >::iterator it=groupSampleMap.begin();it!=groupSampleMap.end();it++){
+ vector<int> indices = it->second;
+ for(int i=0;i<indices.size();i++){
+ for(int j=0;j<i;j++){
+ if(indices[i] > indices[j]) { within += rankMatrix[indices[i]][indices[j]]; }
+ else { within += rankMatrix[indices[j]][indices[i]]; }
+ numWithinComps++;
}
}
-
- out.close();
- }
+ }
- return 0;
+ within /= (float) numWithinComps;
- }
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "driver");
- exit(1);
- }
-}
-//**********************************************************************************************************************
+ double between = 0.0;
+ int numBetweenComps = 0;
-int AnosimCommand::driver(int start, int num, vector<string> names, string pidValue, vector< vector<double> >& completeMatrix) {
- try {
+ map<string, vector<int> >::iterator itB;
- //create a new filename
- ofstream out;
- string outputFile = outputDir + m->getRootName(m->getSimpleName(phylipfile)) + "anosim" + pidValue;
- m->openOutputFileAppend(outputFile, out);
- out.setf(ios::fixed, ios::floatfield); out.setf(ios::showpoint);
- cout.setf(ios::fixed, ios::floatfield); cout.setf(ios::showpoint);
-
- //for each combo
- for (int c = start; c < (start+num); c++) {
-
- if (m->control_pressed) { out.close(); return 0; }
-
- //get set names
- vector<string> setNames;
- for (int j = 0; j < namesOfGroupCombos[c].size(); j++) { setNames.push_back(namesOfGroupCombos[c][j]); }
-
- vector<string> thisCombosSets; //what set each line in the distance matrix is from to be used when calculating SSWithin
- set<int> indexes;
- for (int k = 0; k < names.size(); k++) {
- //is this group for a set we want to compare??
- if (m->inUsersGroups(designMap->getGroup(names[k]), setNames)) {
- thisCombosSets.push_back(designMap->getGroup(names[k]));
- indexes.insert(k); //save indexes of valid rows in matrix for submatrix
- }
- }
-
- int numGroups = thisCombosSets.size();
-
- //calc the distance matrix
- matrix.clear();
- matrix.resize(numGroups);
-
- for (int k = 0; k < matrix.size(); k++) { for (int j = 0; j < matrix.size(); j++) { matrix[k].push_back(1.0); } }
-
- if (thisCombosSets.size() == 0) {
- m->mothurOut("[ERROR]: Missing distance info for sets. Skipping comparison."); m->mothurOutEndLine();
- }else{
-
- if (setNames.size() == 2) { out << setNames[0] << "-" << setNames[1] << '\t'; m->mothurOut(setNames[0] + "-" + setNames[1] + "\t"); }
- else { out << "all" << '\t'; m->mothurOut("all\t"); }
-
- //fill submatrix
- int rowCount = 0;
- for (int j = 0; j < completeMatrix.size(); j++) {
-
- if (indexes.count(j) != 0) { //we want at least part of this row
- int columnCount = 0;
- for (int k = 0; k < completeMatrix[j].size(); k++) {
-
- if (indexes.count(k) != 0) { //we want this distance
- matrix[rowCount][columnCount] = completeMatrix[j][k];
- matrix[columnCount][rowCount] = completeMatrix[j][k];
- columnCount++;
- }
- }
- rowCount++;
- }
+ for(map<string, vector<int> >::iterator itA=groupSampleMap.begin();itA!=groupSampleMap.end();itA++){
+
+ for(int i=0;i<itA->second.size();i++){
+ int A = itA->second[i];
+ map<string, vector<int> >::iterator itB = itA;
+ itB++;
+ for(itB;itB!=groupSampleMap.end();itB++){
+ for(int j=0;j<itB->second.size();j++){
+ int B = itB->second[j];
+ if(A>B) { between += rankMatrix[A][B]; }
+ else { between += rankMatrix[B][A]; }
+ numBetweenComps++;
+ }
}
-
- //calc anosim
- calcAnosim(out, setNames.size(), thisCombosSets);
}
}
- out.close();
-
- return 0;
+ between /= (float) numBetweenComps;
+ double Rvalue = (between - within)/(numSamples * (numSamples-1) / 4.0);
+
+ return Rvalue;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "driver");
+ m->errorOut(e, "AnosimCommand", "calcWithinBetween");
exit(1);
}
}
+
//**********************************************************************************************************************
-int AnosimCommand::calcAnosim(ofstream& out, int numTreatments, vector<string> thisCombosLookupSets) {
+
+vector<vector<double> > AnosimCommand::convertToRanks(vector<vector<double> > dist) {
try {
+ vector<seqDist> cells;
+ vector<vector<double> > ranks = dist;
- vector<seqDist> rankMatrix = convertToRanks();
-
- double meanBetween, meanWithin, RValue, pValue;
-
- meanWithin = calcWithin(rankMatrix, thisCombosLookupSets);
- meanBetween = calcBetween(rankMatrix);
-
- int N = matrix.size();
-
- //calc RValue
- RValue = (meanBetween - meanWithin) / (double) (N * (N-1) / 4.0);
-
- //calc Pvalue
- int count = 0;
- for (int i = 0; i < iters; i++) {
- if (m->control_pressed) { break; }
-
- //randomly shuffle names to randomize the matrix
- vector<string> copyNames = thisCombosLookupSets;
- random_shuffle(copyNames.begin(), copyNames.end());
-
- meanWithin = calcWithin(rankMatrix, thisCombosLookupSets);
- meanBetween = calcBetween(rankMatrix);
-
- //calc RValue
- double randomRValue = (meanBetween - meanWithin) / (double) (N * (N-1) / 4.0);
- if (randomRValue <= RValue) { count++; }
+ for (int i = 0; i < dist.size(); i++) {
+ for (int j = 0; j < i; j++) {
+ if(dist[i][j] != -1){
+ seqDist member(i, j, dist[i][j]);
+ cells.push_back(member);
+ }
+ }
}
- pValue = count / (float) iters;
-
- out << RValue << '\t' << pValue << endl;
- cout << RValue << '\t' << pValue << endl;
- m->mothurOutJustToLog(toString(RValue) + "\t" + toString(pValue) + "\n");
-
- return 0;
- }
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "calcAnisom");
- exit(1);
- }
-}
-//**********************************************************************************************************************
-double AnosimCommand::calcWithin(vector<seqDist>& thisMatrix, vector<string> thisCombosLookupSets) {
- try {
- double within = 0.0;
- int count = 0;
- for (int l = 0; l < thisMatrix.size(); l++) {
- //if you are from the same treatment
- if (thisCombosLookupSets[thisMatrix[l].seq1] == thisCombosLookupSets[thisMatrix[l].seq2]) {
- within += thisMatrix[l].dist; //rank of this distance
- count++;
+ //sort distances
+ sort(cells.begin(), cells.end(), compareSequenceDistance);
+
+ //find ranks of distances
+ int index = 0;
+ int indexSum = 0;
+ for(int i=0;i<cells.size()-1;i++){
+
+ index = i;
+ indexSum = i + 1;
+ while(dist[cells[index].seq1][cells[index].seq2] == dist[cells[index+1].seq1][cells[index+1].seq2]){
+ index++;
+ indexSum += index + 1;
+ }
+
+ if(index == i){
+ ranks[cells[i].seq1][cells[i].seq2] = i+1;
+ }
+ else{
+ double aveIndex = (double)indexSum / (double)(index - i + 1);
+ for(int j=i;j<=index;j++){
+ ranks[cells[j].seq1][cells[j].seq2] = aveIndex;
+ }
+ i = index;
}
}
- within /= (float) count;
-
- return within;
+ if(indexSum == cells.size() - 1){
+ ranks[cells[cells.size()-1].seq1][cells[cells.size()-1].seq2] = indexSum + 1;
+ }
+
+ return ranks;
}
catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "calcWithin");
+ m->errorOut(e, "AnosimCommand", "convertToRanks");
exit(1);
}
}
+
//**********************************************************************************************************************
-double AnosimCommand::calcBetween(vector<seqDist>& thisMatrix) {
- try {
- double total = 0.0;
-
- for (int l = 0; l < thisMatrix.size(); l++) {
- total += thisMatrix[l].dist; //rank of this distance
- }
-
- total /= (float) thisMatrix.size();
-
- return total;
- }
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "calcBetween");
- exit(1);
- }
-}//**********************************************************************************************************************
-vector<seqDist> AnosimCommand::convertToRanks() {
- try {
- vector<seqDist> ranks;
-
- for (int i = 0; i < matrix.size(); i++) {
- for (int j = 0; j < i; j++) {
- seqDist member(i, j, matrix[i][j]);
- ranks.push_back(member);
- }
+
+map<string, vector<int> > AnosimCommand::getRandomizedGroups(map<string, vector<int> > origMapping){
+ try{
+ vector<int> sampleIndices;
+ vector<int> samplesPerGroup;
+
+ map<string, vector<int> >::iterator it;
+ for(it=origMapping.begin();it!=origMapping.end();it++){
+ vector<int> indices = it->second;
+ samplesPerGroup.push_back(indices.size());
+ sampleIndices.insert(sampleIndices.end(), indices.begin(), indices.end());
}
- //sort distances
- sort(ranks.begin(), ranks.end(), compareSequenceDistance);
+ random_shuffle(sampleIndices.begin(), sampleIndices.end());
- //find ranks of distances
- vector<seqDist*> ties;
- int rankTotal = 0;
- for (int j = 0; j < ranks.size(); j++) {
- rankTotal += (j+1);
- ties.push_back(&ranks[j]);
-
- if (j != (ranks.size()-1)) { // you are not the last so you can look ahead
- if (ranks[j].dist != ranks[j+1].dist) { // you are done with ties, rank them and continue
-
- for (int k = 0; k < ties.size(); k++) {
- float thisrank = rankTotal / (float) ties.size();
- (*ties[k]).dist = thisrank;
- }
- ties.clear();
- rankTotal = 0;
- }
- }else { // you are the last one
-
- for (int k = 0; k < ties.size(); k++) {
- float thisrank = rankTotal / (float) ties.size();
- (*ties[k]).dist = thisrank;
- }
+ int index = 0;
+ map<string, vector<int> > randomizedGroups = origMapping;
+ for(it=randomizedGroups.begin();it!=randomizedGroups.end();it++){
+ for(int i=0;i<it->second.size();i++){
+ it->second[i] = sampleIndices[index++];
}
}
- return ranks;
+ return randomizedGroups;
}
- catch(exception& e) {
- m->errorOut(e, "AnosimCommand", "convertToRanks");
+ catch (exception& e) {
+ m->errorOut(e, "AnosimCommand", "randomizeGroups");
exit(1);
}
}
+
//**********************************************************************************************************************
projects / mothur.git / blobdiff